Search Results

ShuntCheck is an aid to the detection of flow in implanted cerebrospinal fluid (CSF) shunts. ShuntCheck includes Micro-Pumper, a component which may be used to temporarily increase CSF flow in suspected non-flowing, patent shunts during the ShuntCheck test. ShuntCheck cannot alone diagnose CSF shunt function or malfunction. The clinical diagnosis of CSF shunt function or malfunction, incorporating the flow information from ShuntCheck, should be made only by a qualified neurosurgeon.

ShuntCheck is a non-invasive device which detects flow in a CSF shunt via transcutaneous thermal dilution. The device consists of a single use disposable thermosensor array patch which is connected to a data acquisition unit (a DAQ) which is connected to a laptop or tablet computer. The device also includes a Micro-Pumper which vibrates the shunt valve during the test procedure to generate a temporary increase in flow in patent but temporarily non-flowing shunts. The shunt is cooled transcutaneously by placing an instant ice pack over the shunt cephalic to the thermosensor. The thermosensor array patch, which is placed on the skin over the shunt "downstream" of the ice, reads the change in skin temperature over the shunt as cooled fluid flows downstream and also at a two nearby control locations. Data is transferred through the DAQ and captured in the computer. If the device detects a characteristic downstream transcutaneous temperature dip, the computer reports "flow confirmed" and presents a time-temperature graph of test data. If no temperature dip is detected, the unit reports "flow not confirmed" and presents a time-temperature graph.

The ShuntCheck III device is intended as an aid to detect flow in implanted cerebrospinal fluid (CSF) shunts. The study presented is a 510(k) submission for substantial equivalence to a predicate device, ShuntCheck v2.2. The evaluation includes both bench testing and limited clinical testing.

Here's a breakdown of the requested information:

1. Table of Acceptance Criteria and Reported Device Performance

2. Sample Size Used for the Test Set and the Data Provenance

• Bench Testing (without Micro-Pumper): The sample size for simulating flow detection (10 ml/hr and 0 ml/hr, threshold, and misalignment) is not explicitly stated as a number of tests or runs. It mentions "bench testing was conducted" and "the ShuntCheck test is conducted normally."
• Bench Testing (with Micro-Pumper):
  • Shunt flow generation and impact on valve function: Testing of eight shunt valves and 5 programmable valves.
  • ShuntCheck's ability to detect flow generated by Micro-Pumper: Not explicitly stated as a number of tests or runs.
• Clinical Testing:
  • 38 asymptomatic patients for testing ShuntCheck without Micro-Pumper.
  • 12 patients (asymptomatic or confirmed functioning shunts) for testing with the proposed Micro-Pumper.
  • 4 patients with confirmed obstructed shunts for testing with the proposed Micro-Pumper.
• Data Provenance: The clinical testing was conducted at Boston Children's Hospital, indicating US-based, prospective clinical data collection for the clinical testing aspects. The bench testing would be lab-based.

3. Number of Experts Used to Establish the Ground Truth for the Test Set and the Qualifications of Those Experts

• For the clinical testing, the ground truth for "shunt function or malfunction" was confirmed via MRI imaging.
• The clinical diagnosis of CSF shunt function or malfunction, incorporating flow information, "should be made only by a qualified neurosurgeon." This implies that a qualified neurosurgeon ultimately interpreted the MRI imaging to establish the ground truth for shunt status in the clinical study. The exact number of neurosurgeons involved in ground truth establishment is not specified. Their qualifications are stated as "qualified neurosurgeon."

4. Adjudication Method for the Test Set

• The document does not explicitly describe an adjudication method for the ground truth (e.g., 2+1 reads, consensus panel). It states that shunt function or malfunction was "confirmed via MRI imaging," implying a standard diagnostic process overseen by medical professionals (qualified neurosurgeons).

5. If a Multi-Reader Multi-Case (MRMC) Comparative Effectiveness Study was done, If so, what was the effect size of how much human readers improve with AI vs without AI assistance

• No MRMC comparative effectiveness study was done. This device is not an AI-assisted diagnostic tool for human readers; it's a standalone device outputting a "Flow Confirmed" or "Flow Not Confirmed" result. The clinical study evaluated the device's accuracy in detecting flow in relation to a patient's shunt status and the impact of the Micro-Pumper component.

6. If a Standalone (i.e. algorithm only without human-in-the-loop performance) was done

• Yes, a standalone performance evaluation was done through both bench testing and clinical testing. The device's software algorithm provides the "Flow Confirmed" or "Flow Not Confirmed" result based on its sensor data, independent of human interpretation of raw data. The "Accuracy without Micro-Pumper" and "Accuracy with proposed Micro-Pumper" in the clinical study table directly reflect the standalone performance against the ground truth.

7. The type of ground truth used (expert consensus, pathology, outcomes data, etc.)

• Bench Testing: The ground truth was established by precisely controlled flow rates (e.g., 10 ml/hr, 0 ml/hr) in a simulated environment using infusion pumps and drop counters.
• Clinical Testing: The ground truth for shunt function or malfunction was established by MRI imaging.

8. The Sample Size for the Training Set

• The document describes a 510(k) submission for a device, not an AI/ML algorithm that typically requires a large, explicitly defined "training set." The device relies on a validated algorithm for detecting flow. The development of this algorithm would have involved various bench and potentially early clinical data, but a specific "training set size" in the context of an AI/ML model is not provided or applicable in the way it's usually understood for deep learning. The testing described serves as validation and verification of the final algorithm's performance.

9. How the Ground Truth for the Training Set Was Established

• Given that this is not an AI/ML device with an explicit training set in the modern sense, the concept of "ground truth for the training set" as it relates to machine learning is not directly applicable. The algorithm's development (and thus its "training" or optimization) would have been based on principles of thermal dilution and fluid dynamics, likely informed by experimental data from bench models. The document refers to the algorithm as a "validated algorithm."

Ask a specific question about this device